open-llm-leaderboard-old/details_adamo1139__Yi-34B-200K-AEZAKMI-RAW-2901
收藏数据集概述
数据集来源
该数据集是在对模型 adamo1139/Yi-34B-200K-AEZAKMI-RAW-2901 进行评估运行期间自动创建的,评估结果展示在 Open LLM Leaderboard 上。
数据集结构
- 配置数量:63个配置,每个配置对应一个评估任务。
- 运行次数:数据集来自1次运行。每个运行结果作为一个特定的分割存储在每个配置中,分割名称使用运行的时间戳。
- 训练分割:"train" 分割始终指向最新的结果。
- 结果配置:一个额外的配置 "results" 存储所有运行的聚合结果,用于计算和显示聚合指标。
数据加载示例
python from datasets import load_dataset data = load_dataset("open-llm-leaderboard/details_adamo1139__Yi-34B-200K-AEZAKMI-RAW-2901", "harness_winogrande_5", split="train")
最新结果
以下是 2024-02-02T03:25:38.031965 运行的最新结果:
python { "all": { "acc": 0.7324210063058885, "acc_stderr": 0.029359152172342876, "acc_norm": 0.7374002642900267, "acc_norm_stderr": 0.029909325332768254, "mc1": 0.39167686658506734, "mc1_stderr": 0.017087795881769625, "mc2": 0.5509212115856775, "mc2_stderr": 0.015260405621718143 }, "harness|arc:challenge|25": { "acc": 0.6296928327645052, "acc_stderr": 0.01411129875167495, "acc_norm": 0.6493174061433447, "acc_norm_stderr": 0.013944635930726099 }, "harness|hellaswag|10": { "acc": 0.6521609241187014, "acc_stderr": 0.0047531124327286995, "acc_norm": 0.8498307110137423, "acc_norm_stderr": 0.0035650718701954478 }, "harness|hendrycksTest-abstract_algebra|5": { "acc": 0.43, "acc_stderr": 0.04975698519562428, "acc_norm": 0.43, "acc_norm_stderr": 0.04975698519562428 }, # 其他任务的结果... }
配置详情
-
config_name: harness_arc_challenge_25
- 分割: 2024_02_02T03_25_38.031965
- 路径:
**/details_harness|arc:challenge|25_2024-02-02T03-25-38.031965.parquet
- 路径:
- 分割: latest
- 路径:
**/details_harness|arc:challenge|25_2024-02-02T03-25-38.031965.parquet
- 路径:
- 分割: 2024_02_02T03_25_38.031965
-
config_name: harness_gsm8k_5
- 分割: 2024_02_02T03_25_38.031965
- 路径:
**/details_harness|gsm8k|5_2024-02-02T03-25-38.031965.parquet
- 路径:
- 分割: latest
- 路径:
**/details_harness|gsm8k|5_2024-02-02T03-25-38.031965.parquet
- 路径:
- 分割: 2024_02_02T03_25_38.031965
-
config_name: harness_hellaswag_10
- 分割: 2024_02_02T03_25_38.031965
- 路径:
**/details_harness|hellaswag|10_2024-02-02T03-25-38.031965.parquet
- 路径:
- 分割: latest
- 路径:
**/details_harness|hellaswag|10_2024-02-02T03-25-38.031965.parquet
- 路径:
- 分割: 2024_02_02T03_25_38.031965
-
config_name: harness_hendrycksTest_5
- 分割: 2024_02_02T03_25_38.031965
- 路径:
**/details_harness|hendrycksTest-abstract_algebra|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-anatomy|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-astronomy|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-business_ethics|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-clinical_knowledge|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-college_biology|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-college_chemistry|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-college_computer_science|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-college_mathematics|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-college_medicine|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-college_physics|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-computer_security|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-conceptual_physics|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-econometrics|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-electrical_engineering|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-elementary_mathematics|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-formal_logic|5_2024-02-02T03-25-38.031965.parquet**/details_harness|hendrycksTest-global_facts|5_2024-02-02T03-25-38.031965.parquet
- 路径:
- 分割: 2024_02_02T03_25_38.031965



